Integrated metabolomic and microbiota analysis of semen: seasonal and morphological associations

Abstract

Recent studies have begun to investigate the metabolic and microbiota profiles in semen, yet their association with abnormal sperm morphology, particularly in teratozoospermia, remains insufficiently characterized. Identifying specific metabolites and microbial taxa linked to this condition could improve diagnostics and management for male infertility. This study analyzed semen samples from 231 patients, including 30 patients with teratozoospermia and 30 patients with normal sperm morphology, collected over four seasons in Chongqing, China. Metabolomic profiling by gas chromatography-mass spectrometry (GC-MS) and microbiota composition analysis via 16S ribosomal ribonucleic acid (rRNA) sequencing revealed distinct seasonal metabolomic shifts, with significant changes in summer and autumn. After excluding seasonally affected metabolites, 14 key metabolites were associated with teratozoospermia, including reduced levels of 4-hydroxyphenylpyruvic acid, phenylpyruvic acid, and N-acetyl-L-aspartic acid. These metabolites are involved in pathways related to oxidative stress and energy metabolism in spermatozoa, suggesting that their depletion may contribute to sperm abnormalities. Proteobacteria, Firmicutes, and Actinobacteriota were predominant phyla across all seasons and groups, but significant genus-level fluctuations, such as Acinetobacter and Staphylococcus, were observed. In teratozoospermia, genera such as Lactobacillus and Limnochordaceae showed differential abundance, correlating with key metabolites and suggesting potential functional interactions. Limnochordaceae showed a significant positive correlation with undecanoic acid, whereas Lactobacillus showed a negative correlation. These findings highlight that while seasonal changes significantly influence semen metabolomics and microbiota composition, teratozoospermia is characterized by specific, season-independent metabolic and microbial signatures. Our study provides insights into the metabolic and microbial dynamics of semen, suggesting the possibility of developing novel diagnostic tools and therapeutic strategies for male infertility.

INTRODUCTION

Semen quality is a critical factor influencing the success rates of assisted reproductive technologies (ART), and it is increasingly recognized as an early sentinel of the organism’s state of health owing to its high sensitivity to environmental changes, oxidative stress, and other exogenous factors.1 This sensitivity makes it an important indicator of overall health, particularly in the context of male fertility. Various parameters, including sperm count, motility, and morphology, are associated with fertilization and clinical pregnancy.2 As male infertility contributes to nearly 50% of all infertility cases, it is crucial to understand the underlying factors that influence semen quality in order to optimize outcomes for couples undergoing ART.

Studies have shown that seasonal variation significantly affects human semen quality, including sperm concentration, total sperm count, motility, and the proportion of normal morphology spermatozoa.3 Sperm concentration and total sperm count are higher in winter and spring, but lower in summer,4 and the percentage of spermatozoa with normal morphology peaks in winter, suggesting that environmental temperature may modulate semen quality.4,5

Advances in high-throughput sequencing have uncovered the role of the male reproductive microbiome in fertility. Semen is not sterile but hosts a unique microbiota, which is closely linked to semen quality and reproductive potential.6 For instance, Lactobacillus is found in higher abundance in normal semen samples, while Prevotella and Pseudomonas are significantly elevated in samples of poorer quality.7 Pathogenic bacteria such as Chlamydia trachomatis, Gardnerella vaginalis, and Ureaplasma species, even in asymptomatic individuals, may reduce fertility by reducing semen quality.8 The composition of the seminal microbiota has been linked to sperm motility and morphology, with Lactobacillus being more prevalent in samples with normal sperm morphology.9

The composition and diversity of seminal microbiota vary significantly between individuals. A study has classified semen into three primary microbiota types: Lactobacillus dominated, Prevotella dominated, and polymicrobial communities, each of which may influence sperm quality and function differently.9 Recent studies using advanced techniques such as metagenomics have advanced our understanding of how microbial communities affect reproductive health.10 However, challenges remain, including inconsistencies in study findings owing to differences in experimental design, sample handling, and data analysis, as well as a limited understanding of the precise mechanisms by which microbiota affect sperm quality and male fertility.

In recent years, the integration of metabolomics into reproductive health research has provided valuable insights into the biochemical factors influencing male fertility, particularly semen quality and sperm morphology. Studies have shown that the composition and abundance of specific metabolites in both spermatozoa and seminal plasma are closely related to various spermiogram profiles. For example, the abundance of specific amino acids, carnitine, and some biogenic amines in seminal plasma are correlated with sperm concentration and morphology, while the abundance of certain amino acids (arginine and glutamine) and taurine in spermatozoa are linked to motility, so they could be potential biomarkers for male infertility.11 Metabolomic profiling has revealed how aging affects the seminal metabolome, with specific metabolites reflecting declines in semen quality and increased sperm DNA fragmentation in older males.12,13 Numerous metabolites, including amino acids and lipids, can be used to differentiate normozoospermic from oligoasthenoteratozoospermic men, uncovering underlying biochemical alterations and pathophysiological mechanisms associated with impaired semen quality.14 This potential of metabolomic analysis to uncover key biomarkers and mechanisms underlying impaired semen quality offers a deeper understanding of male infertility.

Given the known impact of seasonal factors, microbiota, and metabolites on semen quality, this study integrates metabolomic and microbiota analyses to explore associations between these factors and abnormal sperm morphology (teratozoospermia). Specifically, the study aims to identify key metabolites and bacterial genera linked to teratozoospermia, while accounting for seasonal variations.

PARTICIPANTS AND METHODS

Study cohort

A total of 231 men providing semen for in vitro fertilization (IVF) were recruited from the Center for Reproductive Medicine at The Second Affiliated Hospital of Chongqing Medical University (Chongqing, China) between August 2022 and September 2023. Men with a history of smoking, or those diagnosed with diabetes mellitus, hypertension, liver disease, kidney disease, hematological or metabolic disorders, hereditary conditions, sexually transmitted diseases, or genital deformities or dysplasia were excluded from the study. The climate in Chongqing features a gradual increase in temperature from January to June, extreme heat between June and August, and a gradual decrease from September to December. The lowest temperature is in January, with the highest in July and August. Seasonal categories were determined by the date of semen collection: winter (December−February, the average highest temperature [AHT]: 11°C−13°C, n = 33), spring (March−May, AHT: 20°C−27°C, n = 63), summer (June−August, AHT: 30°C−36°C, n = 37), and autumn (September−November, AHT: 17°C−28°C, n = 98), as shown in Supplementary Table 1. Teratozoospermia was defined as having more than 98% of spermatozoa with morphological defects. This threshold exceeds the standard definition outlined by the sixth edition of the World Health Organization (WHO) laboratory manual for the examination and processing of human semen, which defines teratozoospermia as >96% abnormal forms, and was chosen to select for cases with particularly severe morphological abnormalities to enhance the potential contrast with the control group.15 Patients with teratozoospermia (n = 30) were matched with control patients (n = 30, normal morphology >6%) using propensity score matching (Table 1).

Supplementary Table 1.

Sample groups categorized by seasonal variations

	Spring (n=63)	Summer (n=37)	Autumn (n=98)	Winter (n=33)	P
Age (years)	32.38±4.14	32.38±4.14	31.76±3.59	32.63±5.20	0.27a
Duration of infertility (years)	2.00 (1.00–4.00)	2.00 (1.00–3.00)	2.00 (1.00–4.00)	2.00 (1.00–4.00)	0.86b
Previous IVF attempts	1.00 (0.00–1.00)	1.00 (0.00–2.00)	0.00 (0.00–2.00)	0.00 (0.00–1.00)	0.83b
Sperm count (106)	225.00 (180.00–360.00)	192.50 (150.00–320.00)	240.00 (160.00–327.50)	210.00 (150.00–260.00)	0.23b
Semen volume (ml)	3.00 (3.00–3.50)	3.00 (2.00–4.00)	3.00 (2.00–3.88)	3.00 (2.50–3.50)	0.90b
Sperm concentration (106/ml)	80.00 (65.00–110.00)	70.00 (50.00–90.00)	80.00 (60.00–110.00)	75.00 (60.00–85.00)	0.06b
Total motility (%)	70.0 (55.0–75.0)	60.0 (50.0–70.0)	65.0 (46.3–70.0)	60.0 (55.0–65.0)	0.01b
Progressive motility (%)	60.0 (50.0–70.0)	50.0 (40.0–60.0)	55.0 (35.0–60.0)	55.0 (50.0–60.0)	0.00b
Normal morphology (%)	4.0 (2.7–5.0)	4.0 (2.5–4.9)	3.9 (2.5–4.9)	3.8 (2.9–5.2)	0.94b
Abnormal morphology (%)	96.0 (95.0–97.3)	96.0 (95.1–97.5)	96.1 (95.0–97.5)	96.2 (94.7–97.1)	0.94b

^aOne-way ANOVA with Tukey’s post hoc test (normal data). Data presented as mean±s.d; ^bKruskal–Wallis test with Dunn’s post hoc analysis (non-normal data). Data presented as median (IQR) for non-normal variables. IVF: in vitro fertilization; IQR: interquartile range; s.d.: standard deviation

Table 1.

Clinical characteristics of the controls and teratozoospermia patients

Characteristic	Teratozoospermia (n=30)	Control (n=30)	P
Age (year), median (IQR)	32 (28.25–33.75)	32 (30.00–34.75)	0.51a
Duration of infertility (year), median (IQR)	2 (1–3)	3 (1–6.75)	0.18a
Previous IVF attempts, median (IQR)	0 (0–1)	0.5 (0–1)	0.52a
Sperm count (106), mean±s.d.	227.98±115.42	257.95±134.47	0.35b
Semen volume (ml), median (IQR)	3 (3–4)	3 (3–3.5)	0.57a
Sperm concentration (106 ml−1), mean±s.d.	69.30±33.66	80.80±33.63	0.19b
Total motility (%), mean±s.d.	59.2±14.3	63.8±11.0	0.16b
Progressive motility (%), mean±s.d.	50.2±16.4	56.2±12.2	0.11b
Abnormal morphology (%), median (IQR)	98.3 (98.1–99.0)	92.6 (91.8–93.3)	<0.001a
Normal morphology (%), median (IQR)	1.7 (1.0–1.9)	7.4 (6.7–8.2)	<0.001a

^aStatistical comparisons used Mann–Whitney U test (non-normal data). ^bStudent’s t-test (normal data) following Shapiro–Wilk normality assessment, semen parameters were assessed after 30-min liquefaction at 37°C using WHO manual guidelines.15 IVF: in vitro fertilization; IQR: interquartile range; s.d.: standard deviation; WHO: World Health Organization

The study was approved by the Ethics Committees of The Second Affiliated Hospital of Chongqing Medical University (Approved No. 2021-184) and conducted in accordance with the Declaration of Helsinki (2013 revision). Written informed consent was obtained from all participants.

Sample collection and analysis

Semen samples were collected after 3 days of sexual abstinence by masturbation. Before the collection of the sample, individuals received guidance on the necessary procedures to minimize the risk of contamination. This included washing their hands thoroughly with soap two or three times, as well as cleansing the penis with warm soapy water, paying particular attention to the glans and the coronal sulcus. Subsequently, they were to wipe the area with 75% alcohol two or three times. The semen was then ejaculated directly into the semen collection cup (Oosafe, Hingham, MA, USA), ensuring that there was no contact with the sterile inner surface of the container. After 30 min of liquefaction at 37°C, the samples were evaluated manually for volume and examined via optical microscopy (Olympus BX53; Olympus Corporation, Tokyo, Japan) for sperm concentration and morphology, which were analyzed according to the WHO laboratory manual.15 Morphological assessment was performed on Papanicolaou-stained smears at 100× magnification.

Derivatization of extracted metabolites for gas chromatography-mass spectrometry (GC-MS) analysis

Metabolites were extracted in a methyl chloroformate (MCF) derivatization method adapted from a previous study.16 Semen samples (200 μl) were mixed with 68 μl pyridine and 30 μl MCF (Sigma-Aldrich, St. Louis, MO, USA), agitated for 30 s, and treated with 300 μl chloroform and 800 μl of 50 mmol l⁻¹ sodium bicarbonate. Samples were mixed for 10 s and centrifuged at 1500g for 10 min using an Eppendorf Centrifuge 5810 R (Eppendorf AG, Hamburg, Germany). The chloroform layer was collected and prepared for GC-MS analysis.

GC-MS analysis

Derivatized metabolites were analyzed in an Agilent Intuvo 9000 system coupled with an Agilent 5977B mass spectrometer (Agilent Technologies, Santa Clara, CA, USA), operating at an electron impact ionization of 70 eV. The protocol followed our previous work.12 Briefly, the derivatives were injected in pulsed splitless mode at 290°C, with helium as the carrier gas (1.0 ml min⁻¹ flow rate). Separation was achieved using a BD-1701 capillary column (30 m long × 250 μm internal diameter × 0.25 μm film thickness). The oven temperature was programmed to increase from 45°C to 280°C to separate the metabolites. Metabolites were identified by Automated Mass Spectral Deconvolution and Identification System (AMDIS) software (National Institute of Standards and Technology [NIST], Gaithersburg, MD, USA) and matched to an in-house MS library and the NIST mass spectral library. Relative concentrations were normalized to the internal standard (D4-alanine; Sigma-Aldrich) and tissue weight. Targeted metabolites were quantified from calibration curves with corresponding chemical standards.

DNA extraction, polymerase chain reaction (PCR) amplification, and sequencing

Total microbial genomic DNA was extracted from semen samples by using the Pure Cell/Tissue DNA Isolation Mini Kit (DP813-T8; TIANGEN, Beijing, China) according to the manufacturer’s instructions. DNA quality and concentration were measured after 1.0% agarose gel electrophoresis and a NanoDrop® ND-2000 spectrophotometer (Thermo Scientific, Wilmington, DE, USA). The hypervariable V3–V4 regions of the bacterial 16S rRNA gene were amplified with primers 338F (5’-ACTCCTACGGGAGGCAGCAG-3’) and 806R (5’-GGACTACHVGGGTWTCTAAT-3’) in an ABI GeneAmp® 9700 PCR thermocycler (ABI, Foster City, CA, USA). The PCR mixture included 4 μl 5× Fast Pfu buffer (TransGen Biotech, Beijing, China), 2 μl of 2.5 mmol l⁻¹ deoxynucleotide triphosphates (dNTPs), 0.8 μl of each primer (5 μmol l⁻¹), 0.4 μl Fast Pfu polymerase, 10 ng template DNA, and double-distilled water (ddH₂O) to a final volume of 20 μl. PCR conditions included initial denaturation at 95°C for 3 min, followed by 27 cycles of 95°C for 30 s, 55°C for 30 s, 72°C for 45 s, and a final extension at 72°C for 10 min. All samples were amplified in triplicate. PCR products were purified by using the AxyPrep DNA Gel Extraction Kit (Axygen, Union City, CA, USA) and quantified in a Quantus™ Fluorometer (Promega, Madison, WI, USA). Sequencing was performed on an Illumina MiSeq PE300 platform by Majorbio Bio-Pharm Technology Co., Ltd. (Shanghai, China).

Sequencing data analysis

Sequencing reads were filtered for quality and length by using fastp (version 0.19.6; GitHub, https://github.com/OpenGene/fastp, last accessed on 2019 January 16). Taxonomic assignment was performed using the QIIME2 pipeline. Paired-end reads were merged with the Fast Length Adjustment of Short Reads (FLASh) tool and denoised by divisive amplicon denoising algorithm 2 (DADA2). Amplicon sequence variants (ASVs) were taxonomically assigned by aligning representative sequences to the SILVA rRNA database (SSU 138 release; Leibniz Institute DSMZ, Braunschweig, Germany). Linear discriminant analysis (LDA) effect size was used to identify genera with differential abundance between groups (α = 0.05 and LDA score ≥ 3.0).

Statistical analyses

Data are presented as mean ± standard deviation (s.d.) or median (interquartile range [IQR]) for non-normal variables. Normality was assessed using the Shapiro–Wilk test. Comparisons between groups were performed with Student’s t-test or Mann–Whitney U test. P < 0.05 was considered statistically significant. Metabolite pathway analysis and machine learning models were constructed by MetaboAnalyst 6.0 (https://www.metaboanalyst.ca/).

RESULTS

Seasonal shifts in semen metabolomics

Sperm motility was higher in spring and lower in summer and autumn (Supplementary Table 1). To investigate whether there are seasonal variations in semen metabolomics, samples collected over two consecutive years, covering all four seasons, were analyzed. Partial least squares discriminant analysis (PLS-DA) revealed distinct separation of metabolomic profiles between spring and summer, and between summer and autumn (Figure 1 and Table 2). Specifically, 21 metabolites were upregulated and three were downregulated in summer compared with spring (all P < 0.05; Supplementary Table 2). Similarly, 67 metabolites were upregulated and two downregulated in autumn compared with summer (all P < 0.05; Supplementary Table 2). Smaller shifts were observed between autumn and winter (five metabolites), and between winter and spring (seven metabolites), with all P < 0.05 (Supplementary Table 2).

Figure 1.

Partial least squares discriminant analysis (PLS-DA) score plots of metabolite profiles in semen samples from in vitro fertilization (IVF) patients collected across different seasons. Semen samples (n = 231) were collected in Chongqing (temperature range: 11°C–36°C) and analyzed by gas chromatography-mass spectrometry (GC-MS) after MCF derivatization using an Agilent 5977B MS with BD-1701 column (30 m long × 250 µm internal diameter × 0.25 µm film thickness). Metabolites were identified via AMDIS software matched to NIST/SILVA libraries and normalized to D4-alanine internal standard. (a) Comparison between spring and summer. (b) Comparison between summer and autumn. (c) Comparison between autumn and winter. (d) Comparison between winter and spring. Data processed through MetaboAnalyst 6.0. Component 1, component 2, and component 3 represent latent variables (linear combinations of metabolites) that maximize the separation between the predefined groups.

Table 2.

Performance of partial least squares discriminant analysis models in comparing seasonal semen metabolomics

Model	Accuracy	R2	Q2
Spring vs summer	0.95	0.77	0.66
Summer vs autumn	0.96	0.83	0.76
Autumn vs winter	0.83	0.47	0.17
Winter vs spring	0.76	0.52	0.30

Comparing seasonal groups defined by collection date: winter (December–February, n=33); spring (March–May, n=63); summer (June–August, n=37); and autumn (September–November, n=98). R² represents model fit (variance explained); Q² indicates predictive ability through 10-fold cross-validation

Supplementary Table 2.

Metabolomic changes across seasons in semen samples

Differential metabolites	Log2 (fold change)	P
Summer vs Spring
Phthalic acid	1.66	0.00
Pentadecane	0.68	0.00
Phenol	0.42	0.00
Decanoic acid	1.59	0.00
Dodecane	0.32	0.00
Dimethyl phthalate	3.14	0.00
Dibutyl phthalate	−0.31	0.00
Pyruvic acid	0.86	0.00
Undecane	0.68	0.00
3-methyl-2-oxopentanoic acid	0.47	0.00
Tridecane	0.46	0.00
Hexadecane	0.86	0.00
Tetradecane	0.53	0.00
N-acetyl-L-alanine	0.41	0.00
Dodecanoic acid	0.90	0.00
Pimelic acid	0.39	0.00
Hexadec-9-enoate	−0.26	0.00
N-acetyl-L-lysine	0.48	0.01
N-ethyl-2-isopropoxycarbonylazetidine	0.29	0.01
N-acetyl-L-glutamic acid	0.46	0.02
N-acetyl-L-tryptophan	0.34	0.02
Tridecanoic acid	−0.25	0.02
Octanoic acid	0.74	0.02
Picolinic acid	0.24	0.03
Autumn vs Summer
Phthalic acid	1.96	0.00
Undecane	1.45	0.00
Dodecane	0.60	0.00
Pentadecane	0.92	0.00
3-methyl-2-oxopentanoic acid	0.80	0.00
Tetradecane	0.8	0.00
Tridecane	0.62	0.00
Hexadecane	1.06	0.00
Picolinic acid	0.52	0.00
Decanoic acid	1.69	0.00
Phenol	0.43	0.00
Pimelic acid	0.64	0.00
Octanoic acid	1.03	0.00
Dimethyl phthalate	3.18	0.00
Dodecanoic acid	1.09	0.00
Linolelaidic acid	0.48	0.00
gamma-linolenic acid	0.48	0.00
N-acetyl-L-alanine	0.54	0.00
Tetrachlorethane	0.33	0.00
Succinic acid	0.59	0.00
4-methyl-2-oxopentanoic acid	0.38	0.00
N-ethyl-2-isopropoxycarbonylazetidine	0.42	0.00
13-docosenamide	−0.48	0.00
alpha-linolenic acid	0.43	0.00
Hexachlorethane	0.28	0.00
N-acetyl-L-leucine	0.53	0.00
Decane	0.24	0.00
Aspartic acid	0.52	0.00
Valeric acid	0.35	0.00
Isoleucine	0.30	0.00
Butanedioic acid	0.30	0.00
N-acetyl-L-Serine	0.42	0.00
Valine	0.24	0.00
Leucine	0.25	0.00
N-acetyl-L-phenylalanine	0.46	0.00
2-hydroxybutyric acid	0.52	0.00
Lactic acid	0.33	0.00
Pyruvic acid	0.73	0.00
Benzenepropanoic acid	0.28	0.00
Ethanol, 2-butoxy	0.23	0.00
N-acetyl-L-valine	0.40	0.00
Tyrosine	0.24	0.00
Norleucine	0.61	0.00
Phenylalanine	0.31	0.00
Pipecolic acid	0.62	0.00
Proline	0.43	0.00
Threonine	0.31	0.00
N-acetyl-L-glutamic acid	0.50	0.00
Caffeine	0.49	0.00
Glycine	0.28	0.00
N-Acetyl-L-lysine	0.48	0.00
Alanine	0.36	0.00
Sarcosine	0.36	0.00
Tryptophol	0.65	0.00
Methyl 13-methyltetradecanoic acid	0.48	0.00
Hexanoic acid	0.30	0.00
10-Pentadecenoic acid	0.20	0.00
Lysine	0.26	0.00
Oxidized glutathione	0.40	0.00
ɑ.Ketoglutaric acid	0.40	0.00
Glutamic acid	0.25	0.00
Myristic acid	0.42	0.01
Serine	0.24	0.02
Hexadec-9-enoate	−0.16	0.02
L-Hydroxyproline	0.24	0.03
3-Hydroxybutyric acid	0.27	0.03
Docosapentaenoic acid	0.24	0.03
Heneicosanoic acid	0.14	0.04
Tryptophan	0.38	0.04
Autumn vs Winter
Dibutyl phthalate	−0.32	0.00
13-docosenamide	0.42	0.02
Heneicosanoic acid	−0.18	0.02
Pentadecane	−0.30	0.04
Hexadecane	−0.21	0.04
Winter vs Spring
Undecane	−0.59	0.00
Hexachlorethane	−0.38	0.03
Linolelaidic acid	−0.40	0.03
Gamma-linolenic acid	−0.40	0.03
Benzenepropanoic acid	−0.27	0.04
N-acetyl-L-tryptophan	0.51	0.04
13-Docosenamide	0.38	0.04

Metabolomic signatures reflecting sperm morphology

After excluding seasonally affected metabolites, we performed PLS-DA to assess the relationship between semen metabolites and sperm morphology. This analysis revealed a clear separation between normozoospermic and teratozoospermic samples (Figure 2a). Model performance was validated through 10-fold cross-validation, yielding an accuracy of 0.87, R2 of 0.72, and Q2 of 0.48 (Figure 2b). Among the differential metabolites, undecanoic acid was the only one that exhibited higher levels in teratozoospermia. In total, 14 metabolites with variable importance in projection (VIP) scores >1 were identified (Figure 3a), with 4-hydroxyphenylpyruvic acid, phenylpyruvic acid, and N-acetyl-L-aspartic acid standing out with VIP scores >2 and areas under the receiver operating characteristic (ROC) curve (AUC) exceeding 79.0% (Figure 3b–3d).17 These metabolites were significantly reduced in teratozoospermia compared with normozoospermic controls, with fold changes of 0.23, 0.35, and 0.36, respectively. Correlation analysis revealed strong associations among the 14 key metabolites (Figure 3e). Notably, strong correlations were observed between 4-hydroxyphenylpyruvic acid and phenylpyruvic acid, lignoceric acid and behenic acid, as well as itaconic acid and citraconic acid. Enrichment analysis indicated that these metabolites were primarily involved in pathways related to amino acid metabolism, including phenylalanine, tyrosine, and tryptophan biosynthesis, ubiquinone and other terpenoid-quinone biosynthesis, glyoxylate and dicarboxylate metabolism, and arachidonic acid metabolism (Supplementary Figure 1a ^{(94.2KB, tif)} ).18 Significant metabolic pathways associated with teratozoospermia included the metabolism of phenylalanine, histidine, alanine, aspartate, glutamate, arachidonic and linoleic acid, and the citrate cycle (Supplementary Figure 1b ^{(94.2KB, tif)} ).

Figure 2.

Metabolite profile differentiation in semen between the control (normal morphology > 6%, n = 30) and teratozoospermia (>98% abnormal morphology, n = 30) groups. (a) Partial least squares discriminant analysis (PLS-DA) plots showing the distinct metabolite profiles in semen samples from the control and teratozoospermia groups. (b) Assessment of model performance using 10-fold cross-validation to evaluate the reliability of metabolite profile separation between the two groups. Performance refers to the statistical evaluation of the PLS-DA model’s ability to accurately distinguish between the control and teratozoospermia groups based on their semen metabolite profiles.

Figure 3.

Identification of key metabolites with good predictive ability for teratozoospermia. (a) Variable importance in projection (VIP) scores from the PLS-DA model, ranking metabolites by their contribution to distinguishing between the control and teratozoospermia groups. The X-axis represents the calculated VIP score for each metabolite (listed on the Y-axis), indicating its importance to the model’s separation performance. Only metabolites exceeding the VIP threshold of 1.0 are presented. (b) ROC curve for 4-hydroxyphenylpyruvic acid showing the area under the curve (AUC, left panel). The bar plot (right panel) compares the levels of this metabolite between the control and teratozoospermia groups. (c) ROC curve for phenylpyruvic acid showing the AUC (left panel). The bar plot (right panel) compares the levels of this metabolite between the control and teratozoospermia groups. (d) ROC curve for N-acetyl-L-aspartic acid showing the AUC (left panel). The bar plot (right panel) compares the levels of this metabolite between the control and teratozoospermia groups. The optimal cut-off value, sensitivity, specificity, and 95% CI are indicated on the ROC curve in b–d. (e) Pairwise comparison of discrepant metabolites is shown with a color gradient signifying Spearman’s correlation coefficient. ^*P < 0.05; ^**P < 0.01; ^***P < 0.001. PLS-DA: partial least squares discriminant analysis; CI: confidence interval; ROC: receiver operating characteristic.

Seasonal variations in semen microbiota composition

At the phylum level, Proteobacteria, Firmicutes, and Actinobacteriota dominated the semen microbiota, collectively comprising over 75% of the total across all seasons. No significant seasonal differences were observed at the phylum level (Supplementary Figure 2a ^{(100.2KB, tif)} ). However, at the genus level, the composition of the microbiota varied significantly with season (Supplementary Figure 2b ^{(100.2KB, tif)} ); specifically, the relative abundances of Acinetobacter (P < 0.01), Staphylococcus (P = 0.03), Corynebacterium (P < 0.01), and Chloroplast (P < 0.01) fluctuated (Supplementary Figure 2c ^{(100.2KB, tif)} ). Despite these shifts, no significant differences were detected in the alpha diversity indices (Ace richness estimator [P = 0.07] and Shannon diversity index [P = 0.54]) across seasons (Supplementary Figure 3 ^{(50.8KB, tif)} ).19

Interactions between altered semen metabolite composition and microbiota in teratozoospermia

Proteobacteria, Firmicutes, and Actinobacteriota were the predominant phyla in the control and teratozoospermia groups (Figure 4a). At the genus level, the top microbiota were Acinetobacter, Staphylococcus, Lactobacillus, Gardnerella, Corynebacterium, Streptococcus, Limnochordaceae, and Enterococcus (Figure 4b). LDA effect size identified statistically significant genera for each group. After excluding seasonally affected microbiota, Lactobacillus (P < 0.01), Enterobacteriaceae (P = 0.04), Stenotrophomonas (P = 0.04), Massilia (P = 0.03), Pseudoxanthomonas (P = 0.03), and Brevundimonas (P = 0.03) were enriched in controls, while Limnochordaceae (P < 0.01), Fusobacterium (P = 0.03), Sulfurospirillum (P < 0.01), Amaricoccus (P < 0.01), Bacillaceae (P < 0.01), and Peptostreptococcaceae (P < 0.01) were characteristic of teratozoospermia (Figure 4c). Alpha diversity indices (Ace richness estimator [P = 0.98] and Shannon diversity index [P = 0.98]) did not show significant differences between the two groups (Supplementary Figure 4 ^{(51.7KB, tif)} ).

Figure 4.

Bacterial communities in semen microbiota of the control and teratozoospermia groups. Bar charts show mean values of 10 most abundant (a) phyla and (b) genera in the control and teratozoospermia groups. (c) Linear discriminant analysis (LDA) effect size (LEfSe) analysis highlighting differentially abundant genera contributing to group differentiation. The LDA score (shown on the X-axis) represents the magnitude of the difference in abundance of each genus between the groups, with higher absolute values indicating greater contribution to the separation between the control and teratozoospermia groups. Green bars represent genera enriched in the control group, while red bars represent genera enriched in the teratozoospermia group. LEfSe analysis validates the significance of these genera, focusing on those with the highest contribution to group separation.

Cholesteryl benzoate, trans-10-nonadecenoic acid, malic acid, tricosanoic acid, and indole were positively correlated with Lactobacillus, which is probiotic (Supplementary Figure 5 ^{(134.4KB, tif)} ). In contrast, several compounds, including cholesteryl benzoate, trans-10-nonadecenoic acid, tricosanoic acid, indole, 4-hydroxyphenylpyruvic acid, histidine, N-acetyl-L-threonine, and cholesta-3,5-diene, were negatively correlated with Limnochordaceae. Staphylococcus exhibited significant correlations with erucic acid, cholesta-3,5-diene, and cholesta-4,6-dien-3-ol, whereas Streptococcus was significantly correlated with phenylpyruvic acid, glutamine, and docosapentaenoate. Both Phyllobacterium and Limnochordaceae showed significant positive correlations with undecanoic acid, whereas Lactobacillus showed a negative correlation. These associations suggest a potential link between these microbial taxa and seminal plasma metabolite levels in teratozoospermia.

DISCUSSION

This study presents novel insights into how seasonal variations influence both semen metabolomics and microbiota composition, expanding previous research linking environmental factors such as temperature to semen quality and fertility.3,4,20 The findings emphasize the sensitivity of semen as a sentinel of overall health, corroborating its responsiveness to oxidative stress and exogenous factors, as noted in a previous study.1 In this context, it is important to consider the potential effects of environmental pollutants on semen quality and function. These effects include cadmium bioaccumulation, alterations in electrophoretic patterns in acetic acid-urea polyacrylamide (AU-PAGE) and sodium dodecyl sulfate-PAGE, reduced DNA-binding affinity, and the ability to induce oxidative DNA damage.21 It is plausible that similar seasonal variations in pollutant effects could occur in human semen, influencing sperm quality and function.22 Future studies should explore the role of environmental pollutants in shaping the seasonal variations observed in semen metabolomics and microbiota.

The interplay between seasonality, environmental factors, and semen quality may involve multiple mechanisms. Higher temperatures during summer could exacerbate oxidative stress, leading to alterations in metabolites involved in redox homeostasis and energy metabolism, such as glutathione and proline.23 Photoperiod-driven hormonal shifts, including testosterone and melatonin, could further influence spermatogenesis and metabolite levels. Seasonal dietary changes and lifestyle behavior may also indirectly modulate semen metabolomics and microbiota composition. For instance, genera such as Staphylococcus, which show seasonal abundance changes, are known to modulate inflammatory pathways and oxidative stress.24 These microbiota changes, combined with environmental factors, could create a dynamic environment that alters sperm morphology and function.

Seminal plasma contains antioxidants and lipids that protect spermatozoa from oxidative damage and enhance their fertilization potential.25,26 Our observation of reduced antioxidant-related metabolites in teratozoospermia suggests that disruptions in seminal plasma composition may compromise these protective mechanisms, contributing to abnormal sperm morphology. Seminal plasma plays a pivotal role in modulating sperm function through its rich composition of metabolites and microbial communities. Our findings suggest that the reduced levels of metabolites such as 4-hydroxyphenylpyruvic acid and N-acetyl-L-aspartic acid in teratozoospermia may impair seminal plasma’s ability to mitigate oxidative stress, a key factor in maintaining sperm membrane integrity and morphology.27 Meanwhile, the microbiota within seminal plasma, such as Lactobacillus and Limnochordaceae, may shape the seminal environment through metabolite production or consumption, influencing sperm quality indirectly. For example, the positive correlation between Limnochordaceae and undecanoic acid suggests microbial modulation of lipid metabolism, which could affect sperm membrane stability.28 These interactions highlight the dynamic interplay between spermatozoa and seminal plasma, which warrants further mechanistic exploration.

Our study revealed significant seasonal shifts in semen metabolomics and microbiota composition, which are correlated with the observed changes in sperm morphology and motility. The distinct metabolic profiles observed in different seasons, particularly the upregulation of certain metabolites in warmer months, align with the known effects of temperature on sperm quality. Higher temperatures can increase metabolic activity and oxidative stress, leading to alterations in sperm motility and morphology.29 This is further supported by the seasonal variations in sperm characteristics, such as higher motility in spring and lower motility in summer and autumn.30 The interplay between these factors (metabolites, temperature, and microbiota) suggests a complex network of interactions that collectively impact semen quality. This highlights the importance of considering both environmental factors and intrinsic biological processes when interpreting seasonal differences in semen quality and male fertility.

Our analysis revealed that 21 semen metabolites were significantly upregulated in summer compared to spring, while 67 metabolites showed increased levels in autumn relative to summer. These fluctuations suggest that seasonal changes not only affect sperm morphology but also reflect environmental influences (such as temperature), which impact the metabolic composition of seminal plasma and correlate with conditions affecting spermatogenesis and epididymal function. Importantly, the metabolites associated with teratozoospermia highlight the potential of metabolomic profiling as a diagnostic tool for assessing male infertility. Moreover, the correlations found among key metabolites imply a complex network of biochemical interactions that may further affect semen quality and fertility outcome.

Amino acid derivatives, such as N-acetyl-L-aspartic acid, have garnered attention owing to their functional roles in sperm health.31 Our findings reveal lower levels of this metabolite in teratozoospermia samples, which could have implications given its known role in antioxidant activity. As a precursor for glutathione (GSH) synthesis, N-acetyl-L-aspartic acid may help mitigate oxidative stress and inflammation in semen, which are conditions often detrimental to sperm quality and motility.32,33 Other key metabolites, including lignoceric acid and behenic acid, are known long-chain saturated fatty acids that serve as energy reserves and contribute to cell membrane structure and function. Their presence in semen may enhance sperm motility and protect spermatozoa from oxidative damage, reinforcing their potential importance for maintaining cellular integrity.34 These metabolites collectively underline the significant lipid and amino acid metabolism in maintaining semen quality.

Itaconic acid has immunomodulatory effects and inhibits the activity of ten-eleven translocation (TET) dioxygenase, thereby regulating the expression of inflammatory genes.35 Citraconic acid inhibits aconitate decarboxylase 1 (ACOD1) catalysis, reduces interferon responses and oxidative stress, and regulates inflammation and cellular metabolism.36 Decreased levels of these metabolites in teratozoospermia patients imply that immune regulation within semen may play a role in maintaining sperm quality, particularly under conditions of oxidative stress associated with teratozoospermia.

While no significant seasonal differences were observed at the phylum level of the semen microbiota, notable shifts at the genus level were documented. The predominance of Proteobacteria, Firmicutes, and Actinobacteriota across all seasons aligns with previous studies that have identified these phyla as central components of the seminal microbiome.7,8 The seasonal variation in genera such as Acinetobacter and Staphylococcus suggests that environmental factors may also influence microbial diversity within the semen, potentially impacting sperm quality. Interestingly, our findings indicate that certain genera are enriched in teratozoospermia cases, highlighting a potential link between microbiota composition and sperm morphology.

Several researches have revealed associations between certain bacterial genera and semen parameters.7,9,37 In our study, the enrichment of specific genera, such as Lactobacillus in healthy samples and Limnochordaceae and Fusobacterium in teratozoospermia, highlights a potential link between microbiota composition and sperm morphology. A systematic review and meta-analysis shows that the Lactobacillus genus is linked to beneficial semen characteristics in previous studies.38 The positive correlation between cholesteryl benzoate and Lactobacillus suggests a potential role of the latter in supporting sperm health by modulating cholesterol metabolism. Cholesteryl benzoate, integral to cell membrane stability, may contribute to sperm motility and membrane integrity, which are essential for sperm viability and function.28 On the other hand, genera such as Limnochordaceae and Fusobacterium, which are characteristic of teratozoospermia-associated microbiome profile that could contribute to or reflect the abnormal sperm morphology. While the exact mechanisms remain unclear, the associations between microbiota composition and sperm morphology warrant further investigation, as they may offer new insights into microbial contributions to male infertility.

The presence of non-commensal microbes in semen, as identified in our study, may serve as indicators of general health, potentially reflecting reproductive tract infections that impact sperm morphology. Our findings revealed genera like Limnochordaceae enriched in teratozoospermia, suggesting that microbial dysbiosis could contribute to spermatogenic dysfunction in the testes, leading to abnormal sperm forms.39 Furthermore, infections such as epididymitis are often caused by bacteria. They can alter the composition of epididymal fluid, which bathes maturing spermatozoa. This alteration may induce oxidative stress or inflammation, affecting sperm morphology. Infected men typically have lower percentages of morphologically normal sperm.39 A previous study also shows that chronic epididymitis, particularly when combined with elevated leptin, significantly reduces normal morphology rates, highlighting the epididymis as a critical site where microbial presence could influence sperm quality during maturation and storage.40 These insights underscore the interplay between seminal microbiota, reproductive tract health, and fertility outcomes, warranting further investigation into microbial impacts on male reproductive function.

While our study provides valuable insights into the seasonal and morphological associations in semen metabolomics and microbiota, there are several limitations that should be acknowledged. First, the sample size, although substantial, may not be representative of the broader population, particularly in terms of geographic and ethnic diversity. This could limit the generalizability of our findings to other regions and populations. The seminal microbiota, like the gut microbiota, may exhibit substantial geographical variability, influenced by environmental factors such as pollutants, housing conditions, and dietary habits, as highlighted in the study of animal models.10 Multicenter studies involving participants from diverse backgrounds are essential to capture the potential variability in seminal microbiota due to geographical and environmental factors. Second, we did not directly measure the presence of environmental contaminants in the semen samples, which could potentially influence the observed metabolomic and microbiotic profiles. Future studies could incorporate contaminant analysis to better understand their role in shaping semen quality. Additionally, while we accounted for seasonal variations, we did not control for other potential confounding factors such as diet, lifestyle, and occupational exposures, which could also impact semen parameters. Third, we did not explore the influence of host genetics on the seminal microbiota. Evidence from animal studies, such as those in rabbits, suggests that genetic background significantly impacts microbiota composition, with distinct profiles observed among inbred lines.41 Including genetic analyses in future research would help clarify the role of individual genetic variability in shaping microbiota diversity and abundance, further elucidating its impact on fertility outcomes. Finally, the cross-sectional nature of our study limits our ability to establish causality between the observed metabolic and microbial changes and sperm morphology. Longitudinal studies are needed to further elucidate these relationships and their implications for male fertility. In addition, we acknowledge a potential limitation related to the sample processing methodology. In accordance with standard procedures, semen samples were allowed to liquefy at 37°C for 30 min prior to denaturation and subsequent analysis. While liquefaction is a necessary physiological process involving the transformation of gel-like semen into a fluid state, this 30-min incubation period at physiological temperature could potentially allow for the multiplication of bacteria initially present in the ejaculate. Human semen is a complex biological fluid, composed primarily of secretions from the seminal vesicles and prostate gland, which may contain nutrients capable of supporting bacterial growth. Consequently, the microbial profile identified after this ex-vivo incubation might not perfectly reflect the bacterial composition and abundance present at the exact moment of ejaculation. Therefore, while our findings provide valuable insights into the seminal microbiome after standard processing, future studies employing time-course analyses during liquefaction would be beneficial to precisely determine the extent to which this preparatory step influences the detected microbial community. Despite these limitations, our study lays a foundation for future research aimed at understanding the complex interplay between environmental factors, metabolomics, and microbiota in semen.

In summary, this study showed the intricate interactions between environmental factors, metabolomic profiles, and microbiota composition in semen, highlighting their combined influence on male fertility. The seasonally influenced metabolites and genera identified in this research open new avenues for understanding the biochemical and microbial dynamics that support or hinder sperm quality. These findings reinforce the potential of integrating metabolomic and microbiota analyses as diagnostic tools for male infertility, paving the way for targeted interventions that account for environmental influences on reproductive health. Further research is warranted to elucidate the underlying mechanisms linking metabolic shifts and microbial ecology to sperm function, which may yield valuable insights into the management of male reproductive disorders.

AUTHOR CONTRIBUTIONS

SZ conceived and designed the study. JZ, JL, ZY, and YQ performed the experiments. CGS and HZ recruited the patients. JZ, JL, TLH, and SZ analyzed the data and created the figures. SZ and JZ drafted and reviewed the manuscript. All authors read and approved the final manuscript.

COMPETING INTERESTS

All authors declare no competing interests.

Supplementary Figure 1

Pathway analysis comparing the control and teratozoospermia groups. (a) Pathway enrichment analysis conducted to identify significant differences in pathways between the control and teratozoospermia groups. The plot displays enriched pathways, with the Y-axis representing the pathway names and the X-axis showing the enrichment significance as −log₁₀ (P-value), where the P-value indicates the statistical significance of the pathway enrichment based on the differential metabolites identified. (b) Metabolic Pathway Analysis (MetPA) showing the metabolic pathways affected by the differential metabolites identified between the control and teratozoospermia groups. Pathways are visualized with impact scores (indicating the importance of the pathway based on the number and centrality of altered metabolites) and significance as −log₁₀ (P-value). Pathways with P < 0.05 considered significant.

Supplementary Figure 2

Bacterial community composition in semen microbiota across seasons. Bar graphs depict the average relative abundance of the top 10 phyla (a) and genera (b) in semen microbiota samples collected during different seasons. (c) Semen microbiota associated with season showing top 10 genera with significant differences in relative abundance across seasons. P < 0.05; ^**P < 0.01; ^***P < 0.001.

Supplementary Figure 3

Microbial diversity across different seasons. (a) Ace index, a measure of species richness estimating the total number of microbial species (including rare, undetected ones) based on abundance data. (b) Shannon index, a measure of microbial diversity accounting for both species richness and evenness of abundance distribution. Indices were calculated from 16S data after quality filtering with fastp (v0.19.6) and FLASh merging.

Supplementary Figure 4

Microbial diversity between the controls and teratozoospermia patients. (a) Ace index, a measure of species richness estimating the total number of microbial species (including rare, undetected ones) based on abundance data. (b) Shannon index, a measure of microbial diversity accounting for both species richness and evenness of abundance distribution. Indices were calculated based on 16S data from 60 matched samples (30 controls/30 teratozoospermia).

Supplementary Figure 5

Spearman’s rank correlation analysis between the top 20 most abundant microbial taxa and all differential metabolites. Positive correlations are indicated in red, while negative correlations are shown in blue. The intensity of the color corresponds to the magnitude of the correlation coefficient, with darker shades representing stronger correlations. Significance levels are denoted as follows: ^*P < 0.05; ^**P < 0.01; and ^***P < 0.001.